Guidance for Conducting Screening Level Avian Risk Assessments for Spray Applications of Pesticides

By

Douglas J. Urban

Environmental Fate & Effects Division

Office of Pesticide Programs

U. S. Environmental Protection Agency

Washington, D.C.

July 7, 2000

Purpose

This document provides guidance on conducting screening level (preliminary) avian risk assessments for pesticides when they are applied as spray formulations. It covers acute risk and chronic risk, and includes single and multiple applications. The standard risk assessment method used is the Quotient Method which results in Risk Quotients (RQs)[1] -- indices of risk. Screening level RQs are based on appropriate application rates and compared to Levels of Concern (LOCs)[2] [RQs based on other than maximum application rates (e.g., typical, average, most prevalent, etc.) may also be developed and compared to LOCs]. Exposure residues are graphed over time and compared to LOCs to show the estimated duration of risk based on appropriate foliar half-life. These graphs are designed to help the risk managers decide if risk reduction measures should be considered.

The calculated RQs are based on the ecological effects and environmental fate data currently received and evaluated by EFED/OPP as specified in the CFR Part 158 data requirements for registration and reregistration. Screening level first-order decay exposure models such as FATE[3], ELL-Fate and TERREEC are used to estimate pesticide exposure. ELL-Fate or TERREEC would require some modification for the number of days the model is run in order to provide the exposure residues for the graph. The FATE model will be used by way of example throughout the remainder of this guidance.

This guidance will be updated or replaced as additional data and probabilistic assessment tools become available.

Background

The risk quotient methodology described in this guidance has previously been available for both public and scientific review. It has become common in ecological risk assessment to present potential risk in terms of a ratio of the estimated environmental exposure or EEC, divided by the hazard or toxicity such as the LC50, EC50 or NOAEC (No Observed Adverse Effects Concentration). EPA first presented this risk index method in the Standard Evaluation Procedure for Ecological Risk Assessment in 1986 [3]. These ratios are used to express potential acute and chronic risk to birds, wild mammals, fish and aquatic invertebrates.

Ecological risk assessment is an evolving field. EPA sponsors research and works with academia, industry and others in a continuing effort to refine EFED's ecological risk assessment methodologies. Of particular note is the Ecological Committee on FIFRA Risk Assessment Methods (ECOFRAM) which was formed in June 1997 [http://www.epa.gov/oppefed1/ecorisk/index.htm]. Its purpose is to develop tools and processes within the FIFRA framework for predicting the magnitude and probabilities of adverse effects to non-target aquatic and terrestrial species resulting from the introduction of pesticides into their environment. ECOFRAM was convened in response to a review of OPP’s ecological risk assessments and guidelines in May of 1996 by the FIFRA Scientific Advisory Panel (SAP). While recognizing and generally affirming the utility of the current assessment process and methods for screening risk assessment purposes, the SAP noted that OPP has relied on deterministic methods of assessing the ecological effects of pesticides and strongly encouraged OPP to develop and validate tools and methodologies to conduct probabilistic assessments of ecological risk. This resulted in the formation of ECOFRAM.

In June 1999, EPA sponsored a Peer Input Workshop as part of an initiative to revise the ecological assessment process for pesticides. Representatives from industry, government and academia were invited to review the Draft ECOFRAM Aquatic and Terrestrial Reports and make recommendations for improvement. Following that workshop, an EPA Implementation Team was formed and has begun developing a plan to implement probabilistic ecological risk assessments for pesticides. This Team has recently reported their progress to the SAP in early April, 2000. As tools for probabilistic ecological risk assessments become available and are implemented in OPP/EFED, this guidance will be revised and updated.

As noted previously, risk quotients are used to indicate potential ecological risk. EFED recognizes the limitations of the quotient method. A recent listing of limitations is presented in the ECOFRAM Terrestrial Draft Report and includes the following: There is no quantification of the magnitude and probability of adverse effects occurring; There is an increased dependence on expert judgement as the quotient approaches one; Only single point estimates that usually represent the more sensitive or conservative data are used in the estimate; other available data are usually ignored; Because the estimate is conservatively biased, the safety margin may be large. However, the actual size of the safety margin will remain unknown; The method does not account for space or time; Species tested in the laboratory are assumed equal to those in the field; An evaluation of the effect of risk mitigation measures is difficult; It does not account for variability in response from age; It does not account for variance in slope of distribution within and among species; It does not account for variability from environmental conditions; It does not account for effects of short-term, medium-term exposure on sub-lethal endpoints; It does not account for other pathways of exposure and effects such as dermal, inhalation, preening, etc.8

While EFED recognizes that risk quotients generally are not helpful for quantifying risk incrementally, it believes that they can be useful in determining whether risks are likely to be high or low.[4] The ECOFRAM Terrestrial Draft Report noted that quotient values provide a “crude index of magnitude of effects and therefore could be used for comparisons among alternative compounds where comparable data are available.”[5] While the objective of EFED is to move toward probabilistic risk assessment methods, current deterministic methods such as the quotient have not been dismissed. Rather, they remain an integral component of the current risk assessment for the registration and reregistration of pesticides. This is consistent with current Agency guidance for Ecological Risk Assessment4 and with the recent recommendations from the draft ECOFRAM reports for Level 1 or Tier 1 risk assessments. The ECOFRAM Terrestrial Draft Report gives a number of reasons for this: “Quotients may serve as an interim measure that provides a bridge for risk assessors and risk managers between current and new probabilistic risk assessment methods; Quotients remain a primary method within the aquatic ECOFRAM proposal and may continue to be used by EPA risk managers; Terrestrial ECOFRAM has not yet conducted case studies to evaluate the proposed probabilistic methods, therefore it is premature to eliminate deterministic quotients; Quotients may play a role in future evaluations by providing a benchmark to which new methods could be compared; Further evaluation of risk characterization methods and further development of a Levels of Refinement process may demonstrate that quotients serve a useful purpose in determining the applicability of the risk assessment and identification of scenarios of concern (e.g., during the Problem Formulation stage).”[6]

This quotient method is similar to a predictive model. It is based on data inputs such as laboratory eco-toxicity data, fate data from laboratory and/or field studies, computer generated model exposure estimates, and use data from the pesticide labels. The quality of the results from any model reflects the quality of the input data and the adequacy of the models used to accurately represent the most significant processes affecting a pesticide’s fate and biological effects in the environment, and the dependence of those behaviors on the selected input parameters. Nevertheless, EFED believes that risk quotients provide a ‘crude index of magnitude of effects’ i.e., potential risk, that is useful in screening and preliminary risk assessments.

The current methods are intended to be components of a full risk assessment. A complete ecological risk assessment of a pesticide use would include acute and chronic risk quotient calculations for other non-target organisms such as wild mammals, non-target plants, and aquatic organisms, a comprehensive assessment of available field effects data (terrestrial field studies) and incident reports. The risk characterization section of the risk assessment would include a discussion of the risk quotients, the field effects data and incident reports as well as other site specific environmental characteristics that could modify the potential risk indicated by risk quotients.

Avian Effects Data

EPA typically receives the following required laboratory studies to use in performing avian risk assessments: one acute bird LD50[7] (mg/kg) and two LC50[8] (ppm) studies, and two chronic bird reproduction studies, providing a NOAEC[9] (ppm). The preferred test species are the northern bobwhite quail (Colinus virginianus) and the mallard duck (Anas platyrhynchos). Often, and especially for granular formulations, EFED receives LD50 studies on songbirds. EFED evaluates the studies and classifies them as either core[10], supplemental[11] or invalid[12], as well as indicating whether the supplemental and invalid studies are upgradable[13]. The toxicity values from the core and supplemental studies are used in risk assessment.

Avian LD50 (mg/kg), LC50 (ppm) and chronic NOAEC (ppm) values for the most sensitive species tested (the lowest values) are selected from the data submitted to support the pesticide registration. Since relatively few species are used in standard toxicity testing, it is likely that the species most sensitive to each pesticide has not been tested. The few species that are tested often provide a range of toxicity values, reflecting the combined effects of measurement error, variability in sensitivity among individuals within a species, and species-to-species variation in sensitivity to the pesticide being tested. Because of this variation in sensitivity, it is unlikely that this analysis will show the absolute worst case risk for each pesticide considered.

1. Avian Acute Toxicity Endpoints

Based on years of experience in preparing risk assessments, EFED has found that the LD50 value is often a better indicator of acute toxicity to birds than the LC50 value. This seems to be true especially for pesticides with LD50 values less than or equal to 50 mg/kg. Alternately, the LC50 value may be a better indicator of acute toxicity to birds if their LD50 values are greater than 50 mg/kg and they persist in the environment with a foliar half-life greater than one day.

In U.S. EPA 1986 [3], the avian dietary LC50 was presented as the primary acute toxicological endpoint to be compared to the acute exposure. However, Hill [4] points out that oral "ingestion is believed to be the most common route of pesticidal exposure in birds and therefore th[e] oral tests of lethality [LD50 ] provide a sound basis for preliminary screening." Further, he states that "when used in combination and judiciously, the two tests of lethality [LD50 and LC50] are invaluable tools for preliminary evaluation of potential hazard of pesticides to wild birds." Thus, there seems to be a compelling argument to continue requiring both tests and to choose the test results for use in risk calculations based on best professional judgement. Both the avian acute oral LD50 value expressed in mg/kg of body weight[14], and the avian subacute dietary LC50 value expressed in ppm concentration in the diet are used directly in the acute dietary and acute daily ingestion risk calculations, respectively.

2. Avian Chronic Toxicity Endpoint

NOAECs expressed in ppm are typical values resulting from the avian reproduction test. Two species, bobwhite quail and mallard ducks, are tested. Common reproductive effects found in these tests at the Lowest Observed Adverse Effect Level (LOAEC), which is the test concentration just above the NOAEC, are egg thinning, cracked eggs, reduced hatchability, decreased survival rate, reduced growth of F1 generation and reduced egg production. The lowest NOAEC value is used directly in the avian chronic risk quotient calculation.

Estimated Environmental Exposure

Environmental exposure has two components: the frequency and duration of contact with the pesticide; and, the amount or concentration of a pesticide in the environment and available to non-target organisms. The Comparative Analysis of Acute Avian Risk from Granular Pesticides [6] provided an in-depth discussion showing that birds are present in fields treated with pesticides; that the pesticide is available to birds in the fields; and, birds can and do ingest pesticide granules, contaminated plant material, insects, and soil.

The amount of toxicant a bird is likely to consume in the diet or by preening, ingest as a single dose, inhale, or absorb via the eye or through the skin, is currently not well quantified. Only limited data are currently available to determine to what extent ingestion of food items with pesticide residues is incidental, accidental, selected for, avoided or some combination of these possibilities. Research has begun, but is limited at this time [7]. Among bird species, there are tremendous differences in feeding, mating, migration, and other behaviors. Thus, a definitive avian exposure model is not currently available.

Only spray applications are being considered in this guidance. First, EFED uses a simple exposure model to estimate exposure in terms of availability of the pesticide active ingredient on the surfaces of food items likely to be consumed by birds (ppm). Next, EFED attempts to quantify the amount of pesticide likely to be ingested per day by a bird, considering the birds body weight.

EFED views these acute and chronic exposure estimates as generally conservative considering the conservative nature of the model input variables. The actual amount of pesticide available will vary depending on the application rate, application method, configuration and calibration of equipment, wind speed, other field conditions, and the various factors that affect bird ingestion behavior.

  1. Acute Exposure of Birds to Sprayed Pesticides via Diet (ppm available in diet)

In the Standard Evaluation Procedures for Ecological Risk Assessment [3, Table 5], EPA presented a generalized table for estimating pesticide residues on avian food items based on the data compiled by Hoerger and Kenaga [8]. The pesticide residues in the table (all 0-day residues for 1 lb a.i./acre application) have been used to estimate maximum residues likely to be found in avian diets immediately following an application. These estimates have been updated based on Fletcher et al [9] as follows:

Table 1. Estimated Environmental Concentrations (EECs) on Avian Food Items (ppm or mg/kg-diet) Following a Single Application at 1 lb ai/A /
Food Items / EEC (ppm or mg/kg-diet)
Predicted Maximum Residue1 / EEC (ppm) or mg/kg-diet)
Predicted Mean Residue1 /
Short grass / 240 / 85
Tall grass / 110 / 36
Broadleaf/forage plants and small insects / 135 / 45
Fruits, pods, seeds, and large insects / 15 / 7

1 Predicted maximum and mean residues are for a 1 lb ai/a application rate based on Hoerger and Kenaga (1972) as modified by Fletcher et al. (1994).